A brief look at model-based dose calculation principles, practicalities, and promise

Abstract

Model-based dose calculation algorithms (MBDCAs) have recently emerged as potential successors to the highly practical, but sometimes inaccurate TG-43 formalism for brachytherapy treatment planning. So named for their capacity to more accurately calculate dose deposition in a patient using information from medical images, these approaches to solve the linear Boltzmann radiation transport equation include point kernel superposition, the discrete ordinates method, and Monte Carlo simulation. In this overview, we describe three MBDCAs that are commercially available at the present time, and identify guidance from professional societies and the broader peer-reviewed literature intended to facilitate their safe and appropriate use. We also highlight several important considerations to keep in mind when introducing an MBDCA into clinical practice, and look briefly at early applications reported in the literature and selected from our own ongoing work. The enhanced dose calculation accuracy offered by a MBDCA comes at the additional cost of modelling the geometry and material composition of the patient in treatment position (as determined from imaging), and the treatment applicator (as characterized by the vendor). The adequacy of these inputs and of the radiation source model, which needs to be assessed for each treatment site, treatment technique, and radiation source type, determines the accuracy of the resultant dose calculations. Although new challenges associated with their familiarization, commissioning, clinical implementation, and quality assurance exist, MBDCAs clearly afford an opportunity to improve brachytherapy practice, particularly for low-energy sources.

Initial evaluation of Advanced Collapsed cone Engine dose calculations in water medium for I‐125 seeds and COMS eye plaques

“…For the Nucletron selectSeed, dose differences of > 30% were observed beyond the end welds of the seeds between MCNP6, TG‐43, and ACE calculated doses. Average differences in other regions were much smaller, within 1–5% . The minimum radial distance and radial increments for and in OcB were 5 mm and the polar angle increments were 5°.…”

“…. The maximum differences at the outer scleral surface for the model 6711 seed were up to 5 times smaller than for the selectSeed . This is attributed to the lower resolution of the PSS data for the selectSeed compared to the model 6711 seed, as well as to the use of older TG‐43 data which showed poorer agreement with both MCNP6 and ACE.…”

“…The I‐125 brachytherapy seed created for simulations in MCNP6 was the Oncura OncoSeed model 6711 (GE Healthcare Inc., Arlington Heights, IL, USA). The Nucletron selectSeed model 130.002 (Elekta Brachytherapy, Veenendaal, NL, USA) was initially implemented and evaluated in ACE and based on the results obtained, a number of improvements were made for implementation of the model 6711 seed . The results presented here are primarily for the 6711 seed, though references to the selectSeed will be made for comparison.…”

A comparative assessment of inhomogeneity and finite patient dimension effects in 60 Co and 192 Ir high-dose-rate brachytherapy

“…One aspect that has not been systematically reviewed in the literature is the relative importance of patient inhomogeneities and bounded dimensions in 192 Ir and 60 Co HDR brachytherapy. This is timely, since brachytherapy has recently advanced towards individualized planning dosimetry with the commercial availability of TPSs, including image-based dose calculation algorithms as an option for 192 Ir HDR treatments [ 13 , 14 ]. Image-based dose calculation algorithms mark an improvement over TG-43 based planning dosimetry.…”

The effect of density overrides on magnetic resonance-guided radiation therapy planning for lung cancer

“…Model-based dose calculation algorithms include physical models to incorporate the effect of heterogeneities on lateral scattering into dose calculation. Typical examples of model-based dose calculation algorithms include kernel superposition, deterministic solution of the linear Boltzmann transport equation or Monte-Carlo methods [3] , [4] .…”